Giannis Karyotakis

Software & Machine Learning Engineer · MSc Candidate, NTUA

Portrait of Ioannis Karyotakis

Athens, Greece

Hi, I'm Giannis.

I'm a software and machine learning engineer who builds production AI applications, recommendation systems, computational pipelines, and developer tools. I'm currently pursuing an MSc in Data Science & Machine Learning at the National Technical University of Athens (NTUA), and I co-founded EduCoach, an AI platform for career guidance and lifelong learning.

My work sits at the intersection of machine learning, software engineering, and cybersecurity, from HPC performance optimisation on the ARIS supercomputer to empirical studies of how developers work.

Current roles
Co-Founder & Software Engineer · EduCoach
Performance Engineer / Research Assistant · BaLab, AUEB
Languages
Greek (native) · English (C2) · German (B2)

Experience

Co-Founder & Software Engineer

EduCoach · educoach.io

May 2025 – Present
  • Co-founded EduCoach, an AI platform for career guidance and lifelong learning; own the full software lifecycle (architecture, implementation, deployment, maintenance) of a live production system and lead its data engineering.
  • Designed and built a recommendation engine that suggests undergraduate and master's programmes from user-interaction data, using skill-matching algorithms and real-time labour-market data.
  • Won 1st place at Ennovation and a course distinction for the product.

Performance Engineer / Research Assistant

Business Analytics Lab, Athens University of Economics and Business · Part-time, on-site

Mar 2026 – Present
  • Profiling, optimisation, and infrastructure benchmarking of large-scale workloads on High Performance Computing (HPC) systems.
  • Run and manage compute jobs on the ARIS national supercomputer (GRNET) through the SLURM workload manager.
  • Accelerate performance-critical Python code with Numba (JIT compilation) for heavy data-processing pipelines.

Next.js Developer

Dalia Labs

Jul 2025 – Feb 2026
  • Built front-end features for web applications using Next.js and modern JavaScript tooling.

Security Researcher / Software Engineer

SecOPERA · AUEB Research Team

Oct 2024 – Dec 2025
  • Built C and Python tooling for software analysis and hardening of complex open-source software and hardware, using containerisation technologies.
  • Work contributed to peer-reviewed tools on software dependency analysis and software security.

Education

M.Sc. in Data Science & Machine Learning In progress

National Technical University of Athens (NTUA)

9.25/10

Sep 2025 – Jan 2027 (expected)

Thesis (in progress): dynamic resource management of multi-agent AI workflows: serverless-style autoscaling, cold-start handling, and heterogeneous CPU/GPU placement.

B.Sc. in Management Science & Technology

Athens University of Economics and Business (AUEB)

9.27/10

Oct 2021 – Jul 2025

Thesis: Making Android Apps Safer: Improving the Portability of Tools that Monitor In-App Behaviour (SliceDroid). Advisors: Diomidis Spinellis & Nikolaos Alexopoulos.

Thesis PDF

Publications

2026

  • Conference

    An Empirical Comparison of Security and Privacy Characteristics of Android Messaging Apps. PDF
    41st ACM/SIGAPP Symposium on Applied Computing (SAC '26), 2026.

    I. Karyotakis, F. T. Proestakis, E. Talos, D. Spinellis, N. Alexopoulos

    First author. Hybrid static and dynamic analysis methodology, applied to Messenger, Signal, and Telegram, to compare complexity, attack surface, permissions, and network behaviour.

    DOI arXiv
  • Journal

    TGIF: The Evolution of Developer Commit Times.
    Empirical Software Engineering 31:52, 2026.

    I. Karyotakis, E. Talos, D. Spinellis

    First author. Large-scale analysis of GitHub commit timestamps over the last decade and how developer working patterns have evolved.

    DOI

2025

  • Conference, Tool Demo

    PyTrim: A Practical Tool for Reducing Python Dependency Bloat. PDF
    IEEE/ACM ASE 2025, Tool Demos Track.

    K. Karakatsanis, G. Alexopoulos, I. Karyotakis, F. T. Proestakis, E. Talos, P. Louridas, D. Mitropoulos

    Co-author. An end-to-end system that detects and removes unused Python dependencies across source and configuration files, evaluated on 971 open-source packages.

    DOI arXiv

Technical Skills

Languages
  • Python
  • Java
  • C#
  • SQL
  • R
  • JavaScript / TypeScript
  • Unix / Bash
ML & Data
  • Machine Learning
  • Data Science
  • Data Analysis
  • OpenAI API platform
  • LangSmith
Backend / Web
  • Next.js
  • HTML5 / CSS
  • Bootstrap
  • REST APIs
HPC & Cloud
  • ARIS supercomputer (GRNET)
  • SLURM
  • Numba
  • Profiling & optimisation
  • Benchmarking
  • Containerisation
  • Google Cloud Platform
Practices
  • Software analysis & tooling
  • Empirical software engineering
  • Software security

Honors & Awards